Using genome-wide SNP data, we calculated genomic inbreeding coefficients (FROH > 1 Mb, FROH > 2 Mb, FROH > 8 Mb and FROH > 16 Mb) derived from runs of homozygosity (ROH) of different lengths (>1, >2, >8 and > 16 Mb) as well as from levels of homozygosity (FHOM). We compared these values of inbreeding coefficients with those calculated from pedigrees (FPED) of 1422 bulls comprising Brown Swiss (304), Fleckvieh (502), Norwegian Red (499) and Tyrol Grey (117) cattle breeds. For all four breeds, population inbreeding levels estimated by the genomic inbreeding coefficients FROH > 8 Mb and FROH > 16 Mb were similar to the levels estimated from pedigrees. The lowest values were obtained for Fleckvieh (FPED = 0.014, FROH > 8 Mb = 0.019 and FROH > 16 Mb = 0.008); the highest, for Brown Swiss (FPED = 0.048, FROH > 8 Mb = 0.074 and FROH > 16 Mb = 0.037). In contrast, inbreeding estimates based on the genomic coefficients FROH > 1 Mb and FROH > 2 Mb were considerably higher than pedigree-derived estimates. Standard deviations of genomic inbreeding coefficients were, on average, 1.3–1.7-fold higher than those obtained from pedigrees. Pearson correlations between genomic and pedigree inbreeding coefficients ranged from 0.50 to 0.62 in Norwegian Red (lowest correlations) and from 0.64 to 0.72 in Tyrol Grey (highest correlations). We conclude that the proportion of the genome present in ROH provides a good indication of inbreeding levels and that analysis based on ROH length can indicate the relative amounts of autozygosity due to recent and remote ancestors.